DuckDB is a fantastic in-process analytical database, and its Python package is no exception. It’s surprisingly powerful, even on low-resource setups. However, there’s a catch—some basic DuckDB functionality relies on runtime installation of extensions that are downloaded from the internet by default. This design choice seems odd for several reasons.

In many work environments, there are strict controls over what can be called from production or live environments, and external traffic is often blocked. This limitation can hinder DuckDB’s usability in such settings. Setting up an internal repository specifically for DuckDB extensions is just one more hurdle to adoption.

For python ecosystem, we can do better. To address this, I’ve put together a simple repository that publishes these extensions to PyPI, making them easy to install. In my experience, many companies that use python have already whitelisted PyPI via a proxy or maintain an internal mirror for such packages. I hope this approach helps in furthering DuckDB adoption, though I wish this was something supported by the DuckDB maintainers themselves.